Abstract Background The aim of this study was to identify modifiable clinical factors associated with radiographic osteoarthritis progression over 1 to 2 years in people with painful medial knee osteoarthritis. Methods A longitudinal study was conducted within a randomised controlled trial, the “Long-term Evaluation of Glucosamine Sulfate” (LEGS study). Recruitment occurred in 2007–2009, with 1- and 2-year follow-up assessments by blinded assessors. Community-dwelling people with chronic knee pain (≥4/10) and medial tibiofemoral narrowing (but retaining >2mm medial joint space width) on radiographs were recruited. From 605 participants, follow-up data were available for 498 (82%, mean [sd] age 60 [8] years). Risk factors evaluated at baseline were pain, physical function, use of non-steroidal anti-inflammatory drugs (NSAIDs), statin use, not meeting physical activity guidelines, presence of Heberden’s nodes, history of knee surgery/trauma, and manual occupation. Multivariable logistic regression analysis was conducted adjusting for age, sex, obesity, high blood pressure, allocation to glucosamine and chondroitin treatment, and baseline structural disease severity (Kellgren and Lawrence grade, joint space width, and varus alignment). Radiographic osteoarthritis progression was defined as joint space narrowing ≥0.5mm over 1 to 2 years (latest follow-up used where available). Results Radiographic osteoarthritis progression occurred in 58 participants (12%). Clinical factors independently associated with radiographic progression were the use of NSAIDs, adjusted odds ratios (OR) and 95% confidence intervals (CI) 2.05 (95% CI 1.1 to 3.8), and not meeting physical activity guidelines, OR 2.07 (95% CI 0.9 to 4.7). Conclusions Among people with mild radiographic knee osteoarthritis, people who use NSAIDs and/or do not meet physical activity guidelines have a greater risk of radiographic osteoarthritis progression. Trial registration ClinicalTrials.gov, NCT00513422. This original study trial was registered a priori, on August 8, 2007. The current study hypothesis arose before inspection of the data.